Capillary Absorption
Critical Capillary Absorption of Current-Melted Silver
Nanodroplets into Multiwalled Carbon Nanotubes
Yen-Song Chen, Yuan-Chih Chang, Shau-Chieh Wang, Li-Ying Chen, Der-Hsien Lien,
Lih-Juann Chen, and Chia-Seng Chang*
Since the discovery of carbon nanotubes (CNTs),[1] scientists
have intensively investigated their intrinsic characteristics
and tried to explore their properties after combining them
with other materials.[2–15] Due to the large aspect ratio and a
generically uniform diameter, the inner cavity of a CNT has
been used as a nano test tube, siphon, catalyst carrier, and so
on.[4–8] Recently, CNT composites made by filling the CNTs
with various materials were found to enhance the CNTs’ electrical, thermal, optical, and mechanical properties;[9–15] thus,
many methods have been developed to fabricate various CNT
composites. For some CNTs grown with catalysts, such as Cu
and Co9S8, the formation of composites is usually explained
by capillary action.[16–18] Capillarity, by which the liquid is
spontaneously absorbed into the narrow tube, is important for
the control of channel-based devices in some biological and
physical applications. The CNT is one of the smallest capillary
tubes that is known on the nanoscale. Earlier studies suggest
that only low-surface-tension materials can be introduced
into CNTs by capillarity. This nanocapillarity has been held
accountable for the various transition metals encapsulated
inside the CNTs after CNT growth.[11,16] According to recent
theoretical calculations and molecular dynamics simulations,
the single-walled carbon nanotube (SWCNT) with an open
end can act as a “capillary pipette”, which will also absorb
nonwetting metallic nanodroplets.[19–21] Despite all these
theoretical and post-factual findings, no real-time process of
capillary absorption of nonwetting metal droplets by CNTs
has been observed. Herein, we report the real-time capillary action of liquid Ag nanodroplets produced by electriccurrent heating into the CNTs. Our observation reveals that
the capillary absorption of molten Ag droplets can occur only
Y.-S. Chen, Dr. Y.-C. Chang, S.-C. Wang, L.-Y. Chen,
D.-H. Lien, Prof. C.-S. Chang
Institute of Physics, Academia Sinica
Nankang, Taipei 11529, Taiwan
E-mail: [email protected]
Y.-S. Chen, L.-Y. Chen, Prof. C.-S. Chang
Department of Physics
National Taiwan University
Taipei 10617, Taiwan
S.-C. Wang, Prof. L.-J. Chen
Department of Materials Science and Engineering
National Tsing-Hua University
Hsinchu 30013, Taiwan
DOI: 10.1002/smll.201200393
small 2012,
DOI: 10.1002/smll.201200393
when the ratio of the Ag droplet’s size to the inner diameter
of the multiwalled carbon nanotube (MWCNT) goes below a
critical value, and this value decreases as the inner diameter
reduces on the scale of nanometers. This finding is important
for understanding the capillarity at the nanoscale and for
applying it to the fabrication of CNT composites.
Figure 1a shows the transmission electron microscopy
(TEM) image of one Ag nanoparticle caught by the open
end of a tailored MWCNT placed on the scanning tunneling
microscopy (STM) probe. The value of η is defined as the size
ratio of the average Ag nanodroplet diameter Di to the inner
diameter Dt of the MWCNT. A sequence of images recording
the in situ uptake behavior of a Ag droplet for the case of
the initial η equal to 1.62 during electric-current heating
are displayed in Figure 1a–d. The values of electric current
applied at each stage are shown in Figure 1e. The chosen Ag
nanoparticle was brought into contact with the upper carbon
Figure 1. a–d) The dynamical process of a Ag nanodroplet being drawn
into the MWCNT. The direction of current is from the STM probe to
the carbon onion. e) Elapse of the passing current during the heating
process corresponding to (a)–(d). f–h) No absorption occurred even
with the application of mechanical compression on the Ag droplet.
© 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
wileyonlinelibrary.com
1
communications
Y.-S. Chen et al.
onion in Figure 1a. When the applied voltage reached 0.03 V
with a current of 1.54 μA, the Ag nanodroplet deformed and
then partly entered into the MWCNT (Figure 1b). To verify
that the insertion of the droplet was not caused by compression, we pulled the MWCNT containing the Ag droplet
slightly away from the upper carbon onion (Figure 1c),
which resulted in a decrease of current (Figure 1e). When
current injection for heating the Ag nanoparticle increased
to 10.37 μA, corresponding to a current density of 5.87 ×
106 A cm−2, the molten Ag droplet was quickly drawn into
the hollow core of the MWCNT. The Ag droplet was surrounded completely by the MWCNT and, though its shape
changed, its mass remained intact. The apparent gap separating the carbon onion and MWCNT in Figure 1d implies
that the speculated compression between them did not occur
in the Ag absorption. To further corroborate this account, we
chose to intentionally press a larger molten Ag droplet on
a CNT (η ≈ 1.9) as shown in Figure 1f–h, and indeed found
no absorption even if the droplet was severely deformed
(Figure 1h). The compression force can be estimated from
the areal change of the Ag droplet under stress and the surface tension for the Ag droplet, which is 966 mN m−1 at the
melting temperature.[22] With the displacement of the droplet
measured along the pressed direction, that is, 3.6 nm, the calculated force is roughly 9.1 nN.
In the above-mentioned experiment, the absorption of Ag
nanodroplets might have been due to other effects such as:
1) the electromigration effect of electronic flow, which may
promote the motion of the Ag droplet;[2,5,11–13,23] 2) the electrowetting effect, in which the applied voltage changes the
interfacial state between the Ag nanodroplet and the inner
graphite shell of the MWCNT from nonwetting to wetting,
and thus triggers capillary absorption;[24] 3) the thermal gradient force, which induces the motion of the Ag nanodroplet
toward lower-temperature environments;[4,6,23] and 4) thermal
evaporation, which vaporizes the Ag cluster into atoms and
then aggregates them at another place.[23] We will examine all
of these possibilities. The first two effects would be induced
spontaneously in our current-heating experiments.
To determine whether the electromigration effect
dominates, the Ag nanodroplet was readily drawn into
the MWCNT despite application of a reverse current of
−12.06 μA (current density of 3.01 × 107 A cm−2) as displayed
in Figure 2a–c. The Ag-filled CNT structure in Figure 2c was
moved into contact with another Ag nanoparticle on the
larger MWCNT in Figure 2d and, with the current injection
of 11.68 μA (current density of 2.92 × 107 A cm−2), the nanoparticle gradually melted and was drawn quickly into the
MWCNT (Figure 2e). Both forward and reverse currents can
cause the absorption, and the small difference between them
implies that the electromigration effect could only have some
minor contribution towards the absorption of the Ag nanodroplet. In the literature,[25] it was found that a current density
of 2.6 × 108 A cm−2 was needed to initiate the electromigration on a Ag nanobeam, which is about one order larger than
our findings. With our measured current density, the potential electromigration force for each individual Ag atom can
be calculated to be 7.5 pN, which corresponds to a shear
stress of 84 MPa upon the whole Ag droplet.[5] Since we have
2 www.small-journal.com
Figure 2. a–c) Absorption of a Ag nanodroplet by the tailored MWCNT
with a negative current flowing from the CNT substrate to the STM
probe. d) The MWCNT is then moved to another CNT containing two Ag
particles. e,f) Extended absorption of Ag droplets by the MWCNT with
successive uptake of two Ag particles in (d) under a positive heating
current. g,h) Absorption of a Ag nanodroplet situated on a CNT by the
tailored MWCNT with a negative current. The contact angle observed in
(h) remains constant. i) Elapse of the passing current during the heating
process corresponding to (a) to (f).
already argued that the electromigration had a minor contribution in this study, the force responsible for the absorption
observed here should have a magnitude significantly larger
than 84 MPa. As for the electrowetting effect, we observed
in Figure 2g and h that the nonwetting condition (θc > 90°)
of the Ag droplet on another CNT (before absorption) and
inside the CNT (after absorption) remained unchanged while
the voltage as well as current passing through the CNT were
both increased. If the electrowetting effect did occur, it would
be within our detection limit and can thus be considered as
another minor factor in the phenomenon reported herein.
Considering the thermal gradient force, the Ag nanodroplets
should prefer to move away from the hottest point of a CNT,
which is normally found at the contact ends or the midpoint
of the tube. Although it is likely that some temperature gradient is established along the CNT crossing two electrodes,
if the thermal stress played a significant role it would continue to drive the nanodroplet toward the cooler ends. This
is not observed in our experiment where the nanodroplet
tends to halt right after being drawn into the tube, even with
a continual increase of the injection current for more than
20 s (Figure 2e,f) or even longer (Figure 2b,c). The scenario
© 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
small 2012,
DOI: 10.1002/smll.201200393
Capillary Absorption of Current-Melted Silver Nanodroplets
Figure 3. a–c) Absorption process of one Ag droplet contacted by two
MWCNTs with different inner diameters. d–f) Shape variation of the
Ag droplet inside the MWCNT as the current increased. e) The contact
angle θc ≈ 120° at the interface between the molten Ag nanoparticle and
the inner graphite shell of the MWCNT was obtained at currents above
10 μA. f) The Ag droplet inside the MWCNT began to vanish gradually at
a current of about 27 μA.
shown in Figure 2e and f should help explain away the electromigration effect as well. The process of thermal evaporation did occur in our case but at much higher temperature
as shown in Figure 3f. Once it happens, the Ag nanodroplet
will gradually break away and cannot retain its original mass.
Thus, the effects of thermal gradient force and evaporation
contribute very little in the initial stage of the Ag nanodroplet’s uptake by the CNT.
Figure 3a–c shows the absorption result of one Ag nanoparticle 12.17 nm in diameter simultaneously contacting two
different MWCNTs. The inner diameters of thin and thick
MWCNTs are 4.65 and 10.14 nm, respectively, and the corresponding ratios η1 and η2 are 2.62 and 1.20. When the passing
current reached 11.89 μA, the Ag nanodroplet began to
drain into the inner cavity of the thick MWCNT as shown in
Figure 3b. A small drop in current led to incomplete absorption due to a loose contact resulting from the deformation
of the droplet. When the contact resumed and the current
was increased back to 11.79 μA (current density of 6.08 ×
106 A cm−2), the Ag droplet again drained deep into the thick
MWCNT and remained near its open end (see Figure 3c).
These images directly demonstrate that the success or failure
of the absorption is dependent on the value of η. As seen in
Figure 3d and e, we connected the tubes again and continued
to increase the heating current above 10 μA; the contact angle
θc ≈ 120° was then revealed between the molten Ag nanoparticle and the inner graphite shell of the MWCNT.[21,26] With
the amount of current I ≈ 20 μA, twice as large as needed for
absorption, the Ag droplet did not move in any further (see
Figure 3e,f), which once again confirmed that both electromigration and thermal stress played minor roles. Finally, the Ag
droplet evaporated away gradually at the current I ≈ 27 μA
(see Figure 3f). The temperature at which the Ag nanoparticle
melted by resistive heating could not be determined by our
experimental setup. However, the changes in the Ag nanodroplet’s structural phases at each stage of resistive heating
were observed and could be used as a reference. When the
small 2012,
DOI: 10.1002/smll.201200393
Ag nanoparticle melted, absorption by the CNT normally
took place immediately.
In our experimental observations, the absorption mainly
results from the capillary force, and the shear stress generated
by this force should have a magnitude larger than 84 MPa.
Unlike the macroscopic capillarity, which occurs in a tube
only for the wetting liquids with θc < 90°, the capillary action
at the nanometer scale depends on both the pressure difference, ΔPmeniscus, across the meniscus of the Ag nanodroplet in
the MWCNT and the Laplace pressure, ΔPsphere, between the
Ag droplet and the substrate.[27] A sufficiently small nonwetting metallic droplet can thus be drawn into a CNT, and the
condition for the capillary action is defined by ηc, the critical
size ratio of the nanodroplet to the tube inner diameter. The
value of ηc can be singly determined as η = −1/cos θc , which
has been calculated as ≈2 for an unsupported droplet,[19,28]
if the contact angle θc = 120° is given. However, the contact
angle in the energetic term reflects the adhesion energy per
unit area between the Ag droplet and the CNT, which may
vary with the size of the CNT due to its curvature.
We thus performed systematic experiments to test the
dependence of the value ηc on the CNT’s diameter (Dt). As
shown in Figure 4a, with each CNT of fixed inner diameter
Figure 4. a) Distribution of different η ratios versus Dt. The transition
from absorption (squares) to nonabsorption (circles) is observed for a
specific ηc at each fixed diameter of the CNT. The critical absorption
condition for Ag droplets occurs below the boundary ηc, which decreases
as the diameter of the CNT reduces. b) The downward-pointing triangles
represent the observed real-time contact angles in the experiment. The
upward-pointing triangles are derived, using the equation η = − 1 /cosθc,
from the experimental points near the boundary in (a). The line is a
linear fit through the upward-triangle symbols.
© 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.small-journal.com
3
communications
Y.-S. Chen et al.
we can change the value η by approaching various Ag droplets and the transition from absorption to nonabsorption is
observed at a certain critical value (ηc). The distribution of Dt
in our study spans from 2.5 to 8.0 nm. An obvious monotonically declining boundary curve, which separates the absorption from nonabsorption, was observed from the larger CNT
toward the smaller size. The corresponding ηc value at the
boundary ranges from 1.7 to 1.2. This experimental result
clearly unveils the significance of the CNT’s size for the capillary action. Once again, as evident from the absorption of
nonwetting Ag droplets, it demonstrates that on the scale of
several nanometers, the size-dependent nature of the capillary action of CNTs is remarkably robust. The change of the
critical ratio ηc also represents the variation of the contact
angle θc with the CNT’s diameter, and an increase of contact angle suggests the reduction of adhesive energy per unit
area as the CNT becomes smaller. Referring to the boundary
value of ηc in Figure 4a, using the above equation we can
derive the corresponding contact angle (θc) for each CNT diameter, values of which are depicted in Figure 4b as upwardpointing triangles versus Dt−1 and fitted with a straight line.
This trend is expected as a result of curvature and is similar
to the simulation result obtained by Kutana and Giapis,[29]
in which they showed that the contact angle of the mercury
slug in the SWCNT cavity varied linearly with the inverse of
CNT diameter. The measured contact angles of the droplets
inside different CNTs after capillary absorption are marked
in Figure 4b for comparison, and the general trend goes up
with Dt−1. The scattering of observed contact angles is most
likely due to the instability of molten Ag droplets during current injection. However, the data derived from Figure 4a are
less scattered and allude to the origin of the dependence of
the critical ratio ηc on the CNT diameter.
In conclusion, we have presented in situ observations of
the capillary absorption of current-melted Ag nanodroplets
by tailored MWCNTs. As the ratio (η) of a droplet’s size to
the CNT’s inner diameter goes below a critical value, the nonwetting Ag droplet can be drawn into the hollow core of the
CNT through capillary action. We also found that the contact
angle between a Ag droplet and a CNT inner wall increases
when the inner diameter of the CNT gets smaller. This results
in the dependence of the critical absorption value of ηc on
the CNT’s inner diameter (Dt), that is, ηc will rise monotonically with Dt when it is less than ≈8 nm.
Experimental Section
In situ experiments were performed in an ultrahigh-vacuum (UHV)
transmission electron microscope (JEOL-2000V, working at an
accelerating voltage of 200 kV and electron beam current density
of about 100 pA cm−2) and recorded the dynamical process of nonwetting Ag nanodroplets drawn into the hollow cores of MWCNTs.
The UHV-TEM system was equipped with a charge-coupled device
(CCD) camera of 1 million pixels and a TV recorder. They recorded
images of a capillary process at 2 and 30 frames per second,
respectively. First, MWCNTs were prepared on the knife edge of a
gold sheet by electrophoresis at ambient pressure. This gold sheet
was then mounted on the TEM sample holder and placed into a
4 www.small-journal.com
load–lock chamber. After the load–lock chamber was evacuated,
the holder was quickly transferred to the main chamber of the UHVTEM instrument. The TEM system was also equipped with a movable STM probe with which we first chose an individual CNT and
made physical contact to establish a complete electrical circuit for
heating or measurement. We also prepared suitable open-ended
MWCNTs of specific diameters by extracting their inner shells for
approaching different Ag droplets.[30,31] The procedure involved the
deposition of Ag atoms on the MWCNTs by a UHV electron-beam
evaporator.[32] A chosen Ag nanoparticle was then touched with
the tailored open-ended MWCNT. To trigger the capillary action we
needed to melt the Ag particle, which was achieved by applying
a dc current to generate local heating near the particle. Here, we
define the positive (negative) current referring to the STM probe as
the positive (negative) electrode.
Acknowledgements
We are grateful for the support from the National Nanoscience and
Nanotechnology Program of Taiwan under grant number NSC972120-M-001-008.
[1] S. Iijima, Nature 1991, 354, 56.
[2] B. C. Regan, S. Aloni, R. O. Ritchie, U. Dahmen, A. Zettl, Nature
2004, 428, 924.
[3] P. M. F. J. Costa, D. Golberg, M. Mitome, S. Hampel, A. Leonhardt,
B. Buchner, Y. Bando, Nano Lett. 2008, 8, 3120.
[4] L. Dong, X. Tao, L. Zhang, X. Zhang, B. J. Nelson, Nano Lett. 2007,
7, 58.
[5] K. Svensson, H. Olin, E. Olsson, Phys. Rev. Lett. 2004, 93,
145901.
[6] H. A. Zambrano, J. H. Walther, P. Koumoutsakos, I. F. Sbalzarin,
Nano Lett. 2009, 9, 66.
[7] B. M. Kim, S. Qian, H. H. Bau, Nano Lett. 2005, 5, 873.
[8] J. H. Warner, Y. Ito, M. H. Rümmeli, B. Büchner, H. Shinohara,
G. A. D. Briggs, ACS Nano 2009, 3, 3037.
[9] Y. Gao, Y. Bando, Nature 2002, 415, 599.
[10] P. S. Dorozhkin, S. V. Tovstonog, D. Golberg, J. Zhan, Y. Ishikawa,
M. Shiozawa, H. Nakanishi, K. Nakata, Y. Bando, Small 2005, 1,
1088.
[11] D. Golberg, P. M. F. J. Costa, M. Mitome, S. Hampel, D. Haase,
C. Mueller, A. Leonhardt, Y. Bando, Adv. Mater. 2007, 19, 1937.
[12] G. E. Begtrup, W. Gannett, T. D. Yuzvinsky, V. H. Crespi, A. Zettl,
Nano Lett. 2009, 9, 1835.
[13] K. Kim, K. Jensen, A. Zettl, Nano Lett. 2009, 9, 3209.
[14] L. Jankovič, D. Gournis, P. N. Trikalitis, I. Arfaoui, T. Cren,
P. Rudolf, M. H. Sage, T. T. M. Palstra, B. Kooi, J. D. Hosson,
M. A. Karakassides, K. Dimos, A. Moukarika, T. Bakas, Nano Lett.
2006, 6, 1131.
[15] F. Geng, H. Cong, Phys. B 2006, 382, 300.
[16] Q. Zhang, W. Z. Qian, H. Yu, F. Wei, Q. Wen, Appl. Phys. A 2007,
86, 265.
[17] G. Y. Zhang, E. G. Wang, Appl. Phys. Lett. 2003, 82, 1926.
[18] G. Du, W. Li, Y. Liu, J. Phys. Chem. C 2008, 112, 1890.
[19] D. Schebarchov, S. C. Hendy, Nano Lett. 2008, 8, 2253.
[20] D. Schebarchov, S. C. Hendy, Phys. Rev. E 2008, 78, 046309.
[21] K. Edgar, S. C. Hendy, D. Schebarchov, R. D. Tilley, Small 2011, 7,
737.
© 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
small 2012,
DOI: 10.1002/smll.201200393
Capillary Absorption of Current-Melted Silver Nanodroplets
[22] S. Ozawa, K. Morohoshi, T. Hibiya, H. Fukuyama, J. Appl. Phys.
2010, 107, 014910.
[23] J. Zhao, J. Q. Huang, F. Wei, J. Zhu, Nano Lett. 2010, 10, 4309.
[24] J. Y. Chen, A. Kutana, C. P. Collier, K. P. Giapis, Science 2005, 310,
1480.
[25] B. J. Wiley, Z. Wang, J. Wei, Y. Yin, D. H. Cobden, Y. Xia, Nano Lett.
2006, 6, 2273.
[26] J. Lee, T. Tanaka, K. Seo, N. Hirai, J. G. Lee, H. Mori, Rare Metals
2006, 25, 469.
[27] V. C. Holmberg, M. G. Panthani, B. A. Korgel, Science 2009, 326, 405.
[28] G. R. Willmott, C. Neto, S. C. Hendy, Faraday Discuss. 2010, 146,
233.
small 2012,
DOI: 10.1002/smll.201200393
[29] A. Kutana, K. P. Giapis, Phys. Rev. B 2007, 76, 195444.
[30] Y. C. Chang, Y. H. Liaw, Y. S. Huang, T. Hsu, C. S. Chang, T. T. Tsong,
Small 2008, 4, 2195.
[31] J. Y. Huang, S. Chen, S. H. Jo, Z. Wang, D. X. Han, G. Chen,
M. S. Dresselhaus, Z. F. Ren, Phys. Rev. Lett. 2005, 94,
236802.
[32] M. C. Wu, C. L. Li, C. K. Hu, Y. C. Chang, Y. H. Liaw, L. W. Huang,
C. S. Chang, T. T. Tsong, T. Hsu, Phys. Rev. B 2006, 74,
125424.
© 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Received: February 19, 2012
Published online:
www.small-journal.com
5